>please pardon me for jumping in with an unrelated question. Could this delayed garbage collection that you describe, be the reason why the "right way" to release an object, is to first set the reference to .null. and then release it? IOW, could it be that setting the reference to .null. forces VFP to do this garbage collection?

Garbage collection is triggered on sys(1104) or
if vfp enters a wait state, for example a doevents or wait wind [timeout]

Any difference in the above would only be in the set of objects waiting for the next collection.